How the projections work

The planning tools turn your balances and a handful of assumptions into a year-by-year picture of the next few decades. That picture is a useful way to compare choices. It is not a forecast, and this page is the honest account of why.

The one thing that matters most

Every figure is downstream of numbers you entered: your balances, your expected return, inflation, spending, savings, retirement age, and any life events. We do not check any of them against reality. A projection is arithmetic applied to your guesses — change the expected return by a single percentage point and a 30-year result can move by hundreds of thousands of dollars. Treat the difference between two scenarios as far more meaningful than either scenario's absolute number.

The scenario page's "What moves it" table is the same projection run again with one input changed — spending or savings by 10%, or retirement two years later — and nothing else. Each row shows the FIRE year, whether the money lasts, and what would be left in the final year in today’s dollars against the headline’s own figure. It exists so you can see for yourself which of your two numbers the plan hangs on; for most households it is spending.

What the model does

It steps forward one year at a time. The first year starts from your newest balances, so it covers only the months left until December: September balances get three months of growth, saving and spending, not a whole year again. A life event dated this year is still charged in full, since the app can't tell whether it's been paid yet. Balances from an earlier year are treated as current. While you're working it adds your savings and grows every balance. From retirement on, it grows balances, applies any Social Security and part-time income, adds any pre-Medicare healthcare cost you entered until age 65, and withdraws whatever spending isn't covered. Taxable money goes first. Then Roth, except that traditional money you can already take without penalty goes before Roth. Then any traditional money that is still penalised, and finally the HSA from 65. Traditional withdrawals are grossed up for tax, plus a 10% penalty when you're too young to take them cleanly.

Life events go one of three ways. Most are costs, paid from your savings while you're working and from your accounts once you've retired. A spending drop, such as a paid-off mortgage or the kids leaving home, takes its amount off what you spend in each retired year it covers. It changes nothing while you're still working: your savings figure is what you save, and the model doesn't assume you'd save the money a drop frees up. A drop with no end year also lowers the FIRE, Barista and Coast targets from the year it starts. A drop that ends doesn't, because 25 times a saving that doesn't last would overstate it. Money in, such as an inheritance or the proceeds of a home sale, is added to taxable savings in the year it arrives, without tax. If it lands in the years before penalty-free access, it counts toward the early-retirement bridge. An inherited traditional IRA is taxed when you draw it, so enter what it would be worth after tax.

Your accounts are sorted into buckets, because where money sits changes when you can spend it: taxable (cash, brokerage, bonds, company stock, crypto, other assets), traditional (traditional IRA, 401k, 403b, 457b, SEP, SIMPLE, annuities, pension), Roth, HSA, 529 and custodial, and debts. A traditional account you've marked as your final employer's plan is held separately, because the rule of 55 can only reach that one. Illiquid holdings are excluded entirely — home equity, vehicles, a business, collectibles, a policy's cash value. They're real wealth, but you can't spend a kitchen.

What it deliberately leaves out

These are known simplifications, not bugs. Each one is a place where the projection will differ from your real life:

Taxes

• By default there is one flat effective tax rate on traditional withdrawals, the one you entered. That default has no brackets, no standard deduction, no state tax and no required minimum distributions, and the order you draw from accounts barely changes your tax.
• The "Federal brackets" tax model is closer, but still only federal and simplified. Each retired year runs through the 2026 federal brackets and standard deduction, grown each year by your inflation assumption (the IRS uses a slightly slower index). Tax is charged on traditional withdrawals, part-time pay and 85% of Social Security, plus the state rate you enter, applied flat to the same taxable income. Required minimum distributions start at 73 or 75, depending on your birth year. Any required amount you didn't already spend is taxed and moved into your taxable savings. Left out: the extra deduction for people over 65, the lower Social Security inclusion below the income thresholds, capital-gains tax, the Medicare premium surcharge (IRMAA) and ACA subsidy cliffs, and a surviving spouse moving to the single brackets. The working years still use one rate: the rate the brackets would charge on a year of your spending drawn entirely from traditional accounts. The FIRE, Coast and Barista checks use that same rate.
• The withdrawal order is one of two fixed rules, not an optimizer. "Taxable first" spends taxable savings before traditional money. "Fill the 12% bracket first" draws traditional money each year up to the top of the 12% bracket, then taxable, then Roth. The plan page shows both side by side.
• Roth conversions are a fixed rule too, and leave out health-insurance subsidies. With "Federal brackets" on, a scenario can convert traditional money to Roth each year from retirement until required minimum distributions start, up to the top of the 12% or 22% bracket, after the year's spending. The tax is paid from taxable savings, so once those run out only the amount the standard deduction covers is converted. Before 59½ each conversion is held back for five years, as the law requires, unless spending it early with the 10% penalty is the only option left. Only the converted amount waits; growth on it is treated like the rest of your Roth. Every converted dollar is income for ACA premium subsidies and Medicare's income surcharge (IRMAA), and neither is modeled. If you'll rely on a subsidy before 65, conversions can cost far more than the tax the plan shows.
• Taxable savings are taxed at nothing. Withdrawals from a brokerage account carry no capital-gains tax, and neither does its growth. Under either tax model this makes spending taxable money look cheaper than it is, which slightly favours "Taxable first" in the comparison.
• The FIRE, Coast and Barista milestones allow for the early-withdrawal penalty, but only roughly. Retiring before 60 means some traditional money may have to come out early, and the 10% penalty on it is subtracted before a milestone counts as reached. The model assumes your taxable and Roth money is spent first, charges the penalty only on the traditional money still needed before 60, and ignores growth over those years. It doesn't count ways around the penalty, such as substantially equal periodic payments, or a Roth conversion ladder, even when the scenario runs one. The projection itself and the bridge check do count the ladder.
• The rule of 55 covers one plan. It frees only the accounts you've marked as your final employer's plan, and only if you retire in or after the year you turn 55. It never frees an IRA, including money you've rolled into one. The model can't tell whose plan it is, so if two of you retire in different years, it treats the whole household as retiring at the scenario's age. How the rule of 55 works.
• Roth contribution basis isn't tracked. The model treats Roth balances as available without penalty. In reality only your contributions come out freely before 59½; earnings generally don't.

Markets and returns

• One blended return applies to every bucket. Your cash, your 401k, and your crypto all grow at the same rate. No asset allocation, no glide path, no rebalancing, no fees, and no difference between accounts.
• The main projection has no market crashes. It grows at your assumed rate every single year. Real sequences don't, and the order in which good and bad years arrive matters enormously near retirement.

Income and spending

• Spending is one flat number that rises with inflation, apart from any spending drops you add. Real retirement spending usually isn't flat, and this is the assumption people are most often wrong about.
• Healthcare before Medicare is one flat figure you supply. If you enter a yearly cost for coverage between retirement and 65, the projection charges it in every retired year before 65, grown at its own medical-inflation rate, and counts it in the early-retirement bridge check. Leave it at zero and it isn't modeled at all — the plan page says so when you retire before 65. Premium subsidies, the income cliffs that go with them, HSA reimbursement of premiums and late-life long-term care are not modeled.
• Savings and salary simply grow with inflation. No raises, promotions, job losses, or career breaks.
• Social Security is a single household benefit. One number, adjusted for your claiming age and grown with inflation as a stand-in for COLA, with a fixed share treated as taxable. No spousal or survivor benefits, no earnings record, and no assumption about future changes to the program. There is no earnings test either: a benefit claimed before full retirement age while you're still working is counted and saved in full. A salary would usually see most of it withheld, and credited back later as a higher benefit. The plan page says how much this counts when a scenario claims before it retires.
• Debts are held flat. They don't accrue interest and aren't paid down on a schedule; they simply reduce what you're worth. To model the end of a loan's payments, such as a mortgage paid off, add a spending drop in the year they stop.

About the likelihood percentage

Scenario cards show something like "78% · Likely" under "chance money lasts." Here is exactly what that number is: the model runs 500 simulated lifetimes, varying the annual return randomly around your assumed rate, and reports the share in which the money didn't run out before the end age.

• The volatility is assumed, not yours. One fixed figure stands in for an asset allocation we don't know. It sits between an all-stock portfolio and a balanced one. If your real allocation is different — and it is — the percentage is measuring someone else's risk.
• It inherits every simplification above. Randomising the return does not add taxes, RMDs, healthcare costs, or spending changes to the model.
• Two digits of precision are not two digits of accuracy. The underlying estimate is worth roughly ±2 points at best, before you account for the model being wrong in the ways this page describes. The number is stable between page loads by design, which makes it look more authoritative than it is.
• It is not the probability that your retirement works. It is the share of runs that succeeded inside this model.

Today's dollars

Most headline figures are shown in today's dollars — deflated so that a number thirty years out means what it would buy now. This is the honest way to read a long projection, but it means the figures won't match a future account statement, which will show larger nominal amounts.

The projections on the Goals screen

A goal is an amount by a date, and it carries its own growth rate — the one you typed into the goal, not a figure the app picked for you and not the rate in your plan scenario. Change it and every number on that goal's card moves.

It also carries a basis — what the goal counts. That can be your whole net worth (asset-backed or cash-based), a slice of the portfolio (everything invested for the long term, just the retirement accounts, just the taxable ones, or just cash and savings), or a single earmarked pot (an HSA, a 529, a custodial account). A narrow basis counts only the accounts of that type: a 529 goal reads your 529 balances and nothing else, and it reads zero until you have one. Goals that count down — paying a mortgage off, clearing a card — are not supported; every figure on the card assumes the balance is climbing toward the target.

• The progress bar is not a projection. It is what you have today divided by the target, and nothing else. Two goals for the same amount will show the same percentage however far apart their dates are — that is the bar working correctly, not ignoring the deadline. The dates move the projection below it.
• Growth alone compounds today's balance at your rate to the target date, assuming you never add another dollar.
• Needs $X/mo in new savings is what you would have to contribute each month, on top of that growth, to land exactly on the target. It reads zero once growth alone would clear it.
• Recent pace is how much the balance actually changed per month over roughly the last year. The app records balances, not transactions, so it cannot see which part of that change was money you saved and which part was the market moving. It estimates the saving half by subtracting the growth your own rate implies over that period. If your rate is wrong, that estimate is wrong with it — and a year that grew less than your assumed rate will show a negative savings pace even though you were saving.
• Projected / on track carries growth and that estimated savings pace forward together. One recent year decides the savings half, so a year with a bonus in it, a house sale, or a bad market will swing the verdict.

The rate applies to everything the goal's basis counts. On an asset-backed goal that includes your home and your cars, which do not grow like a stock portfolio — the app does not split them out, so the blended figure is yours to estimate. Entering a stock-market return on an asset-backed goal will project your house appreciating like the S&P 500. A narrower basis is easier to rate honestly, because one number is then describing one kind of holding.

The rate is also flat and certain: the same return every year, no volatility, no bad decade, no sequence risk. Real markets do not deliver an average annually, and the Goals screen — unlike the plan's likelihood percentage — does not model the range.

The coverage figures on the Protect screen

The Protect screen sizes four amounts — an emergency fund, life insurance, long-term disability coverage, and an umbrella liability limit. These are a different kind of number from the retirement projection: no model, no compounding, no decades. Each is a published rule of thumb applied to figures you already entered, and each one on screen states the inputs it used.

• Emergency fund — three to six months of the spending in your plan scenario, against the balance of your checking, savings and cash accounts. Your plan holds total spending rather than essentials only, so the target errs high.
• Life insurance — the DIME method: debts, plus income replacement to the year your youngest turns 18 (ten years when no child under 18 is on file), plus mortgage, plus college still to fund for children who haven't finished it, less the savings your survivors would already have. Shown beside the simpler 10×-income rule as a cross-check.
• Long-term disability — 60% of gross income, as an annual benefit.
• Umbrella liability — your net worth rounded up to a whole million, with a $1M floor.

What they leave out: your health, your state's law, what an insurer would actually charge or approve you for, the terms inside a policy you already hold (whether a disability benefit is taxable, when a term policy expires), and whether any of it suits your circumstances. A rule of thumb is a starting point for a conversation with a licensed agent or adviser, not a substitute for one. Where a figure can't be worked out — no income on file, no plan scenario — the screen says so instead of guessing.

Coverage you haven't recorded shows as nothing recorded, never as none. The app can't know what policies you hold.

What these numbers are good for

Comparing scenarios against each other. Seeing whether retiring at 55 instead of 60 moves things a little or a lot. Noticing that you have a gap before penalty-free retirement-account access. Finding out that your plan is sensitive to one assumption.

What they are not good for: deciding you can retire on a specific date because a specific number appeared. For that, take these scenarios to a qualified financial professional who can see your whole situation — including everything listed above that this model ignores.

This page changes

We improve the model over time, and when we do we update this page. A figure you saw before may differ afterwards. We don't keep or reproduce old versions of a projection.